#asahi · 2021-10-08 16:11
marcanhah#asahi · 2021-10-08 16:10
marcan:)#asahi · 2021-10-08 16:10
marcanj_ey: pcie DT is blocked on that ;)#asahi · 2021-10-08 16:08
marcan*then* next week I should finally have time :p#asahi · 2021-10-08 16:08
marcanand going to spend a bunch of tomorrow practicing tunes for a jazz jam on sunday#asahi · 2021-10-08 16:07
marcanmarcan is going to sleep tonight#asahi · 2021-10-08 16:06
marcanmaz: word#asahi · 2021-10-08 16:05
marcanif somewhere else in Apple someone is thinking of an Apple Silicon Mac Pro...#asahi · 2021-10-08 16:05
marcanj_ey: we know they share these blocks across chips#asahi · 2021-10-08 16:05
marcanthat was a notably missing one in AIC#asahi · 2021-10-08 16:05
marcane.g. priority#asahi · 2021-10-08 16:05
marcanthough it's possible they added features too#asahi · 2021-10-08 16:04
marcanwhich is the max AIC can support#asahi · 2021-10-08 16:04
marcanbecause I bet somewhere someone at Apple is thinking of >31 core chips#asahi · 2021-10-08 16:04
marcanand if I had to guess what they changed in AIC... 64-bit or more CPU affinity registers.#asahi · 2021-10-08 16:03
marcanlol, did they decide to change everything after M1? though that's not terribly surprising, if M1 is the beta version#asahi · 2021-10-08 16:03
marcanI guess we'll see what M1X/M2 do#asahi · 2021-10-08 16:03
marcanwell so much for forwards compat for that one :p#asahi · 2021-10-08 16:03
marcanheh#asahi · 2021-10-08 16:02
marcanit's just their way of shoving init into macos but driving it from iBoot#asahi · 2021-10-08 16:02
marcanI know those are pretty dumb#asahi · 2021-10-08 16:02
marcanyeah#asahi · 2021-10-08 16:02
marcanOTOH that is less generic re: future SoCs#asahi · 2021-10-08 16:02
marcanif we know what the registers are, a better binding would be to explicitly list the reg vals#asahi · 2021-10-08 16:01
marcanthat said, Apple's idea of "tunables" is "register write dumps"#asahi · 2021-10-08 16:01
marcancan't think of another way#asahi · 2021-10-08 16:01
marcanwell, that's going to be one big turd to go into the DT then#asahi · 2021-10-08 16:01
marcandynamic, derp#asahi · 2021-10-08 15:59
marcangot the prop names?#asahi · 2021-10-08 15:59
marcanthe template or dynamic stuff iBoot inserts#asahi · 2021-10-08 15:59
marcanyes but which adt#asahi · 2021-10-08 15:59
marcanif they're from syscfg there's no way around them having to be in the DT then#asahi · 2021-10-08 15:59
marcando we know if they're static or from syscfg yet?#asahi · 2021-10-08 15:58
marcanso we're going to have to do something about those tunables :(#asahi · 2021-10-08 15:58
marcansven: yeah, I think I saw that#asahi · 2021-10-08 15:58
marcanalso need to add irqs to i2c at some point (sven didn't do that yet IIRC)#asahi · 2021-10-08 15:58
marcanyeah, spi is going to be easy anyway#asahi · 2021-10-08 15:57
marcanyeah#asahi · 2021-10-08 15:56
marcanwell, given that the "CFG_DONE" in their GPIO driver turned out to be "INPUT_BUFFER_ENABLE", let's just say I have zero faith in their bit defs#asahi · 2021-10-08 15:56
marcan(that's how I know they're not cleanrooming this either)#asahi · 2021-10-08 15:56
marcanyeah, I do know their took the names from the kexts#asahi · 2021-10-08 15:55
marcanj_ey: cool :)#asahi · 2021-10-08 15:55
marcanwell, I didn't even *look* at their ATC driver (other than that time I opened their... clock driver... that actually starts an ASC?!... and I closed the window in horror) but given what I've heard from people who have... I'm almost certainly better off doing a cleanroom RE from scratch#asahi · 2021-10-08 15:52
marcan:p#asahi · 2021-10-08 15:51
marcanI've long had the "they aren't reverse engineering things properly, or actively obfuscating their sources" spidey sense tickling#asahi · 2021-10-08 15:51
marcanthere's a reason I didn't want to use their code from the start, and it's not because I hate them or love NIH :p#asahi · 2021-10-08 15:50
marcaneither way I will probably rewrite that SPI driver, because let's face it, everyone who tried to upstream a corellium driver so far ended up rewriting most of it anyway#asahi · 2021-10-08 15:48
marcanjust need to spend some time flipping bits#asahi · 2021-10-08 15:48
marcanhowever, still not close enough to share a driver#asahi · 2021-10-08 15:48
marcanbut I suspect some of its antics are just poor reverse engineering, and actually match the Samsung SPI control reg a bit better